Alternative Title
Course Correction: Steering Large Language Models (LLMs) Toward Educational Goals
Contributor
University of Central Florida. Faculty Center for Teaching and Learning; University of Central Florida. Division of Digital Learning; Teaching and Learning with AI Conference (2025 : Orlando, Fla.)
Location
Seminole A
Start Date
29-5-2025 3:15 PM
End Date
29-5-2025 3:40 PM
Publisher
University of Central Florida Libraries
Keywords:
Language education; LLMs; Finetuning; Prompting strategies; AI in education
Subjects
Japanese language--Computer-assisted instruction; English language--Study and teaching--Japanese speakers; Language and languages--Study and teaching--Technological innovations; Applied linguistics--Research; Second language acquisition--Computer-assisted instruction
Description
When designing learning materials and exercises for world language education, precise control over grammar and vocabulary is crucial, especially at the beginner level. However, the output from base LLMs does not match the structured introduction of this content in language curricula, posing a challenge for educators. This presentation explores practical strategies for steering LLMs for use in language education. We'll discuss practical challenges and solutions we've discovered while developing finetuning and prompting strategies for our applications. Though developed for Japanese, these methods have the potential to transfer to other world languages, offering insights into AI' s role in diverse language education.
Language
eng
Type
Presentation
Format
application/pdf
Rights Statement
All Rights Reserved
Audience
Faculty
Recommended Citation
Hollinghead, Alex and Matsui, Hisae, "Course Correction: Steering LLMs Toward Educational Goals" (2025). Teaching and Learning with AI Conference Presentations. 124.
https://stars.library.ucf.edu/teachwithai/2025/thursday/124
Course Correction: Steering LLMs Toward Educational Goals
Seminole A
When designing learning materials and exercises for world language education, precise control over grammar and vocabulary is crucial, especially at the beginner level. However, the output from base LLMs does not match the structured introduction of this content in language curricula, posing a challenge for educators. This presentation explores practical strategies for steering LLMs for use in language education. We'll discuss practical challenges and solutions we've discovered while developing finetuning and prompting strategies for our applications. Though developed for Japanese, these methods have the potential to transfer to other world languages, offering insights into AI' s role in diverse language education.